The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of arbitrary shell commands in the background using the bg run and bg restart commands. This is the primary and documented function of the skill.
[EXTERNAL_DOWNLOADS]: Recommends the installation of the agentcli-helpers package using the uv tool manager. This package provides the underlying bg command-line utility and is a vendor-provided resource for the skill.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface when the agent reads or logs the output of background processes. If a background job produces malicious text, the agent might interpret it as instructions. 1. Ingestion points: Command output read through bg read and bg logs. 2. Boundary markers: No explicit markers are defined to isolate job output from the agent's control flow. 3. Capability inventory: Includes full shell command execution (bg run). 4. Sanitization: No sanitization of process output is performed before presentation to the agent.

bg-jobs